[R] Gamma distribution parameter estimation

Rolf Turner rolf.turner at xtra.co.nz
Mon Aug 8 03:47:28 CEST 2011


On 08/08/11 13:20, Peter Ehlers wrote:
> On 2011-08-07 16:57, Rolf Turner wrote:
>>
>>
>> Neither Ted Harding's post,  nor Prof. Ripley's, really appear to
>> address the OP's follow-up question which was
>> *why* did he get the ``weird error'' (non-finite finite difference, NaNs
>> produced) when he applied fitdistr(),
>> as he was advised to do by Jorge Ivan Velez.  The ``weird error'' did
>> not occur for Jorge, nor did it occur
>> for my very good self.  This presumably has something to do with
>> operating system and/or version of R used,
>> details of which the OP unfortunately did not provide.
>
> Rolf,
>
> It's not the OS or R version. Close inspection of the error msg
> shows that the OP (without mentioning that salient fact) used a
> different set of values to fit.

Oh for Pete's sake!  I just looked at the first entry of the data set (which
is the same as in the original post) and thought that all entries were the
same.  Duhhhh!  Thanks for pointing this out.

     cheers,

         Rolf



More information about the R-help mailing list